KR101584829B1 - Recovery of transmission errors - Google Patents

Abstract

CLAIMS What is claimed is: 1. A method of recovering transmission errors, comprising: receiving a data packet comprising an error detection code (51) associated with data contained in a data packet (50) - checking the error detection code of the received packet to detect an errored state of the associated data, and - if an error condition is detected, determining for each candidate value of the set Determining a finite set of candidate values for key data; determining a marginal likelihood of the candidate values via ancillary data as a function of an error detection code of the received packet; Determining a first correlation between the primary data of the packet and the candidate value, and determining a corrected value for the primary data among the candidate value set as a function of the marginal availability and the first correlation .

Description

METHOD FOR RECOVERING TRANSMISSION ERRORS, APPARATUS FOR RECOVERING TRANSMISSION ERRORS AND RECEPTION OF TRANSMISSION ERRORS

Field of the Invention The present invention relates to the field of data transmission and to the restoration of transmission errors, and more particularly to the use of protocol data units corresponding to a protocol or a stack of protocol layers.

Most networking communication processes are modeled as layers. This layered representation derives the term protocol stack, which refers to a hierarchical stack of protocol suites. By dividing the communication process into layers, the protocol stack allows for partitioning of labor, e. G., Ease of implementation through software, ease of software code testing, and the ability to develop alternative layer implementations. The layers communicate with the upper or lower layer through a concise interface. In this regard, a layer provides services for the immediately above layer and uses the services provided by the layer immediately below it, e.g., a transport service. At each layer, the protocol refers to the set of rules needed to provide the services of the layer.

In a communication network, network devices implementing a given protocol layer transmit and receive data in the form of PDUs (Protocol Data Units). The coding syntax of the PDU is an attribute of the protocol layer. The PDU includes control data and service data of the protocol layer. The service data is client data, i.e., data received from the interface of the protocol layer and used by the service of the layer. Often, the service data is referred to as a payload. The control data is data necessary for specifying and controlling the service provided by the protocol layer. Typically, the control data is placed in the header of the PDU.

According to a first object, an embodiment of the present invention provides a method for recovering a transmission error, the method comprising the steps of: receiving a protocol data unit corresponding to a protocol layer, the format of the protocol data unit comprising at least At least one error detection field for an error detection code associated with the control data, and at least one service field for the service data; and means for detecting an error in the control data Checking the error detection code of the protocol data unit, determining a finite set of candidate values for the control data and determining an error detection code value associated with the candidate value of the set if an error condition is detected, The candidate value for the control data is determined as an information function outside the protocol data unit, Wherein the sub-information includes prior knowledge of the control data; determining a first correlation between the control data of the received protocol data unit and each candidate value; Determining a second correlation between the error detection code associated with the candidate value and selecting a corrected value for the control data from the set of candidate values as a function of the first and second correlations.

In a particular embodiment, the step of determining a candidate value comprises determining a single candidate value for a first portion of the control field and a plurality of candidate values for a second portion of the control field, And combining the candidate value with each candidate value of the second portion. In one embodiment, the first portion of the control field comprises one or more sub-fields of the control field. These subfields may refer to known or predictable fields. In an embodiment, the second portion of the control field comprises one or more subfields of the control field. These subfields may be referred to as unknown fields.

According to a first object, an embodiment of the present invention also provides an apparatus for restoring control data of a protocol data unit corresponding to a protocol layer, wherein the format of the protocol data unit includes at least one control field for control data, At least one error detection field for the error detection code associated with the service data and at least one service field for the service data, the apparatus comprising input means for receiving the protocol data unit, And a control data correction means operable to determine a finite set of candidate values for the control data, wherein the candidate values for the control data are selected from the group consisting of: As a function of information outside the protocol data unit And the external information includes previous information on the control data to determine an error detection code value associated with the candidate value of the set and to determine a first correlation between the control data of the received protocol data unit and each candidate value Determining a second correlation between an error detection code of a received protocol data unit and an error detection code value associated with each candidate value and determining a second correlation between a corrected Select a value.

According to a second object, an embodiment of the present invention provides a method for recovering transmission errors, the method comprising: receiving a data packet including an error detection code associated with data contained in a data packet, - checking the error detection code of the received packet to detect an errored state of the associated data, and - if an error condition is detected, Determining, for each candidate value, a finite set of candidate values for key data; and determining a marginal likelihood of the candidate values via the assistance data as a function of the error detection code of the received packet Determining a first correlation between the primary data and the candidate value of the received packet; And a step of selecting a corrected value for the primary data among the set of values.

This method can be used to correct an error if an error is found in the main data and it is not necessary to attempt to correct the error if an error is found in the ancillary data. In an embodiment, the primary data includes data that is more important, higher priority, or more reliable than the secondary data. This method can be applied to the protocol layer decoder to recover the control data required for the processing of the PDU, and the service data is treated as auxiliary data.

In one embodiment, determining the peripheral likelihood of the candidate value comprises determining a finite set of potential values for the ancillary data, classifying the potential value of the ancillary data into a subset associated with each error detection code value All of the potential values of the ancillary data in the subset are selected to yield an associated error detection code value when combined with the candidate values of the primary data; and for each subset, the ancillary data of the received packet and the received packet Determining a peripheral likelihood of ancillary data belonging to the subset as a function of the error detection code of the subset; and determining the peripheral likelihood of the candidate value by accumulating the peripheral likelihood for all the subset.

In one embodiment, determining the perimeter likelihood of ancillary data belonging to a subset includes determining a second correlation between an error detection code value associated with the subset and an error detection code of the received packet, Determining a third correlation between the latency of the ancillary data in the set and the ancillary data of the received packet and determining the periverage availability of the ancillary data pertaining to the subset as a function of the second and third correlations .

In an advantageous embodiment, the step of determining the peripheral likelihood of the candidate value comprises the steps of: dividing the error detection code of the received packet into a plurality of blocks, and for each block of error detection codes of the received packet, Determining a partial peripheral likelihood of the candidate value as a function of the partial peripheral likelihood associated with all blocks of the error detection code;

In one embodiment, the step of determining the partial peripheral likelihood of the candidate value comprises the steps of: determining a finite set of potential values for the ancillary data; determining a potential value of the ancillary data, Classifying all of the potential values of the ancillary data in the subset into a subset, when the potential values of the ancillary data in the subset are combined with the candidate values of the primary data to yield an associated block value; and, for each subset, Determining a partial peripheral likelihood of ancillary data belonging to a subset as a function of a block of error detection codes of the packet; and determining the partial peripheral likelihood of the candidate value by accumulating partial peripheral likelihoods for all subsets .

In one embodiment, determining the partial perimeter likelihood of ancillary data belonging to a subset includes determining a second correlation between a block value associated with the subset and a block of error detection codes of the received packet, Determining a second correlation between the latent value of the ancillary data in the subset and the ancillary data of the received packet; and determining a partial perihelability of the ancillary data belonging to the subset as a function of the second and third correlations .

In one embodiment, classifying the potential value of the ancillary data into a subset associated with each error detection code value or each block value comprises configuring a trellis representation of the potential value of the ancillary data.

In one embodiment, the trellis representation consists of a lower depth value to a higher depth value, and the depth of the trellis represents the number of bits of the already determined ancillary data.

In another embodiment, the trellis representation consists of a high depth value to a low depth value, and the depth of the trellis represents the number of bits of the already determined ancillary data.

In one embodiment, the same set of auxiliary data potential values is used for all candidate values of primary data. This set can then be determined only once. In another embodiment, another set of ancillary data latent values may be used for other candidate values of the primary data.

In one embodiment, a finite set of potential values for ancillary data includes all possible combinations of binary values for each and every data bit of a field of a packet reserved for ancillary data.

In one embodiment, the packet comprises a protocol data unit, wherein the main data comprises control data of the protocol data and the auxiliary data comprises service data of the protocol data unit. The method may include processing the service data of the received protocol data unit as a function of the corrected value of the control data.

In another embodiment, the packet comprises video data coded with a scalable video codec, wherein the main data comprises video data corresponding to a base layer and the auxiliary data comprises video data corresponding to an enhancement layer.

According to a second object, there is provided an apparatus for recovering transmission errors, comprising: input means for receiving a data packet including an error detection code associated with data contained in a data packet; An error detection code checking means for checking an error detection code of a received packet to detect an errored state of the associated data; a determination unit for determining a finite set of candidate values for the main data, Determining, for the candidate value, the perimeter likelihood of the candidate value via the ancillary data as a function of the error detection code of the received packet, determining a first correlation between the primary data of the received packet and the candidate value, Key data that may be operable to select a corrected value for key data from a set of candidate values as a function of a first correlation And correction means.

According to a third object, an embodiment of the present invention provides a method of transmitting video data, the method comprising encoding a picture sequence with a scalable video codec to generate a first bitstream corresponding to a base layer and a second bitstream corresponding to a enhancement layer Generating a second bitstream; and generating a sequence of data packets, each packet comprising: main data belonging to a first bit stream; auxiliary data belonging to a second bit stream; and an error detection code And transmitting the data packet sequence over a communication channel, e.g., using any suitable transport protocol or protocol stack.

Dependencies define another embodiment of the present invention. Other embodiments are obtained from combinations of claims.

Embodiments of the present invention are based on the idea of using error detection codes to detect and recover transmission errors in selected portions of a PDU or packet and ignoring residual errors affecting other portions of the PDU or packet. Embodiments of the present invention can be applied to a transparent protocol layer decoder capable of transmitting a payload of a PDU irrespective of its effectiveness.

These and other features of the present invention will become apparent with reference to the embodiments described hereinafter for illustrative purposes, and reference is made to the drawings.
1 is a schematic representation of a decoder module according to an embodiment of the present invention.
Figure 2 is a schematic representation of a method of operating a transparent protocol layer that may be executed by the decoder module of Figure 1;
Figure 3 illustrates an embodiment of a protocol stack and corresponding encapsulation scheme in which the method of Figure 2 may be used.
Figure 4 is a schematic representation of a communication network in which the protocol stack of Figure 3 can be used.
Figure 5 illustrates a data receiver in accordance with an embodiment of the present invention that may be used in the communication network of Figure 4;
Figures 6 and 7 illustrate packet formats that correspond to the standard 802.11 PHY and MAC protocol layers, respectively.
Figure 8 illustrates a transmission method that may be used in the communication network of Figure 4;
Figure 9 is a schematic representation of a method for operating a transparent PHY protocol layer in accordance with an embodiment of the present invention.
10 is a graph illustrating the operation of a PHY layer decoder according to an embodiment of the present invention.
11 is a graph illustrating the operation of a MAC-Lite layer decoder according to an embodiment of the present invention.
Figures 12 and 13 are trellis diagrams.
14 is a graph showing the operation of an embodiment of a MAC layer decoder.
15 is a schematic representation of a transmission system according to one embodiment.
Figure 16 illustrates one embodiment of a decoder module that may be used in the transmission system of Figure 15;
17 is a schematic representation of a method of combining a transparent PHY protocol layer and a transparent MAC protocol layer according to an embodiment of the present invention.

1 is a schematic representation of a decoder module 10 for decoding protocol data units corresponding to a predefined protocol. Decoder module 10 is designated for use in the receiver, which receives data encoded according to a corresponding protocol stack from a transmitter via a protocol under consideration or communication channel. In the protocol under consideration, the format of the protocol data unit includes at least one control field for control data, at least one error detection field for an error detection code associated with all or a portion of the control data, And at least one service field. In one embodiment, the PDU is a packet and the control field is a header of the packet.

The error detection code may be any or all of the content of the control field, that is, any type of redundant information computed at the transmitter from the data to be covered, to allow detection of transmission errors at the receiver. In one embodiment, the error detection code is a checksum or a CRC (Cyclic Redundancy Check) or FEC (Forward Error Correction code).

The decoder module 10 comprises an input module 1 for receiving protocol data units. Protocol data units may be obtained in a variety of ways depending on the protocol layer under consideration. In one embodiment, for example, if corresponding to a physical layer decoder, the input module receives a PDU from a detection module (not shown), which scans the received data sequence to detect the PDU. The PDU can be recognized by detecting a predefined data sequence, e.g., a known preamble. In another embodiment, the input module receives a PDU from another decoder module corresponding to a lower protocol layer.

After receiving the PDU, the input module 1 identifies the control field and the error detection field according to the predefined PDU format and forwards the contents of these fields to the error detection code-checking module 2, The integrity of the corresponding control data is checked by using the error detection code of FIG. If the error detection code check is successful, that is, there is no transmission error, the control data of the protocol data unit is assumed to be correct. The PDU is then passed to the processing module 3, which processes the PDU according to the protocol's functionality and control data of the PDU.

The function of the processing module 3 depends on the service implemented by the protocol layer. The function of the processing module 3 is to reorder several PDUs, for example as a function of the sequence number contained in the control data, to distinguish and decapsulate the service data as a function of, for example, the length included in the control data , Collecting and chaining the service data of the multiple PDUs, splitting the service data of the PDUs, and selecting the destination to which the service data should be output for the protocol demultiplexing as a function of the protocol identifier included in the control data . Other types of services will be apparent to those skilled in the art. The processing performed by the processing module 3 may be transmitted in whole or in part, or occasionally in a cascaded or segmented form, to the output 4 of the decoder module 10. For example, in one embodiment where the received data corresponds to a protocol stack, the output 4 may send the service data of the PDU to an upper protocol layer decoder and the processing module 3 may be adapted to the higher protocol layer decoder Service data. Although one output is shown, the decoder module 10 may include several outputs, for example, to output different types of payloads. The processing module may select the output as a function of the control data of the PDU.

In one embodiment, the processing module 3 stores a selected portion of the control data of the PDU to be processed, e.g., a selected sub-field of the header, in the caching module 5 and uses this data in intra- , Which will be described later.

If the error detection code check performed by the error detection code checking module 2 is unsuccessful, that is, if there is a transmission error in the control data of the PDU, the PDU sends a control data modification module 6, Lt; / RTI > The control data modification module 6 determines a set of finite candidate values for the control data with errors and determines an error detection code value associated with the candidate values of the set. Then, the control data modification module 6 uses the error detection code of the received PDU to select an optimal candidate from the previously determined candidate values. To this end, it is necessary, on the one hand, to compare the control data and the error detection code of the received protocol data unit, on the one hand, to compare each candidate value and the associated error detection code value and to determine a set of candidate values, And selects a modified value for the control data from the correlation between the data.

Once the modified value of the control data is selected, the control data modification module 6 sends the modified PDU to the processing module 3. The processing module 3 operates in the manner described above, as a function of the modified control data of the protocol data units in this case.

The set of candidate values may simply be regarded as an exclusive list of all possible combinations of binary values for each and every data bit of the control field. This leads to a computational complexity of approximately 2 ^** l, where l represents the length of the control field in bits and ^{/ ** /} represents the power.

In a preferred embodiment, the control data modification module 6 constitutes a more limited set of candidate values, depending on the previous information on the control data, i. E. The information outside the detected protocol data unit. Depending on the control field, this prior information may be obtained from different sources, such as the protocol layer specification, intra-layer or inter-layer redundancy, or from the context of the communication over which the data sequence is received.

A redundancy in a hierarchy is defined by a certain deterministic relationship between the various identified information items transmitted in the same protocol layer, for example, between the fields of a given PDU and / or between fields of a PDU that is successfully transmitted in the same protocol layer Refers to the existence of a relationship. The redundancy in the hierarchy is used to determine a candidate value or set of candidate values for the control data of the received PDU as a function of the data received in the control field of the previous PDU, Function.

Inter-layer redundancy is defined as a certain deterministic relationship between the various identified information items transmitted in different protocol layers, such as a relationship between one or more fields of a PDU in a first protocol layer and one or more fields in a PDU in a second protocol layer ≪ / RTI > The inter-layer redundancy is used to determine a candidate value or set of candidate values for control data of a received protocol data unit at a given protocol layer as a function of data received at a protocol data unit, in particular at different, upper, lower, or lower protocol layers Function. To implement intra-layer redundancy, the control data modification module 6 may retrieve data from the caching module 5. To implement inter-layer redundancy techniques, the control data modification module 6 may retrieve data from similar caching modules at the upper or lower protocol layer, which is indicated by arrows 7 and 8.

The communication context refers to a known parameter through a communication session between the transmitter and the receiver. These parameters may be fixed at the transmitter and at the receiver, or may be negotiated during the initial phase of establishment of the communication channel, e.g., during the attachment process between the wireless terminal and the wireless base station or access point. The static or dynamic context parameters may also be transmitted in a signaling message sent by a wireless base station or an access point. In one embodiment, the control data modification module 6 may retrieve data from the memory 9 of the receiver for which such parameters were stored, for example, in a configuration file. As a result, candidate values for the subfields corresponding to these parameters can be determined by accessing the memory 9.

In one embodiment, the control data modification module 6 incorporates the specifications of the protocol layer under consideration to maximally reduce the set of candidate values for the control field of the PDU. For example, the specifications of the protocol layer may provide predefined constant values for the subfields of the control field. Therefore, the control data modification module 6 considers only candidate values in which this subfield is set to a predefined value. For other subfields, the specification of the protocol layer may provide a limited list of possible values. Therefore, the control data modification module 6 considers only candidate values in which this subfield is set to a value selected from this limited list. The expression "limited" herein means that the list contains less than all combinations of binary values for each and every data bit of the subfields.

In one embodiment, the control data modification module 6 relies on intra-layer redundancy to determine candidate values for the sub-fields of the control field, which are relatively static parameters, i.e., from one received PDU to the next Lt; / RTI > In addition, intra-hierarchy redundancy can be used to determine candidate values for sub-fields that include parameters that develop in the next predefined manner, e.g., from a received PDU, by increasing or decreasing previously received values . For this subfield, the control data modification module 6 determines a candidate value for the subfield of the protocol data unit as a function of the control data arranged in the subfield of the previously detected protocol data unit.

In some embodiments, the protocol stack may include a plurality of fields in the same or different protocol layers, including correlated information items via similar or pre-defined correlation rules. In a corresponding embodiment, the control data modification module 6 determines one or more candidate values for the subfields of the control field of the PDU using this predefined correlation rule in combination with intra-layer or inter-layer redundancy techniques. For example, the PDU in the first protocol layer encapsulates the PDU in the second protocol layer, i.e., the second layer PDU in the service field. The format of the PDU in the second layer includes a second sub-field indicating the length of the service data and a second sub-field indicating the transmission speed of the service data. The format of the PDU in the second layer includes a second sub-field indicating a period for transmitting the next layer PDU. Thus, the time period received in one second layer PDU allows determination of the relationship between the length of the next second layer PDU to be received and the value of the first field of the PDU encapsulating it. Other examples of inter-layer redundancy, for example, similar information items that are redundantly transmitted in different layers, can be found in the IP / UDP / RTP protocol stack.

In another embodiment, the service field in the first layer is used to convey exactly one PDU in the second layer. Thus, a similar relationship is obtained between the first subfield and the second subfield of a subsequent PDU to be received at the first layer. By retaining only candidate values that satisfy this relationship, the control data modification module 6 can greatly reduce the set of candidate values to be considered for modifying the control data of the subsequent PDU if an error occurs.

In the preferred embodiment, the service data of the PDU is not checked for transmission errors by the decoder module 10, as opposed to control data. Thus, the decoder module 10 implements a so-called transparent protocol layer that can output the payload of a PDU if it is modified or damaged. By allowing the correction of the erroneous control data of the received PDU, the decoder core 10 can interpret and process the upper portion of the PDU received at the corresponding protocol layer, so that the quality of the service data, To reach the next protocol layer or application layer. Preferably, the decoder module 10 receives the service data of the protocol data unit as soft information and processes it under the form thereof to output the soft information corresponding to the service data to the next protocol layer or application layer. Here, the soft information refers to a sampled signal including several levels between an 'L' level indicating a logic '0' and an 'H' level indicating a logic '1'. In contrast, hard information refers to a sampled signal quantified only in terms of logical '0' and '1'. Delivering the soft information of the service data to the next layer enables the use of efficient decoding techniques or error correction techniques in the next layer. In one embodiment, application data in the form of soft information is passed through one or more transparent protocol layers of the receiver to an application layer implementing a common source-channel decoding technique, which results in many errors being corrected. A variety of robust joint source-channel decoders are known, for example, in "On Variable Length Codes for Iterative Source / Channel Decoding" by R. Bauer and J. Hagenauer, 1998, UT, Snowbird, Proceedings of DCC 272-282 .

In the preferred embodiment, the decoder module 10 receives part or all of the PDU in the form of soft information and the control data correction module 6 uses the soft information of the control data to select the best match candidate value. In this embodiment, the control data correction module 6 calculates the correlation with the control data and the error detection code of the received protocol data unit provided as soft information. Using soft information in the calculation improves the accuracy of the selection step.

In a modified embodiment, the error detection code checking module is suppressed and the control data correction module 6 receives all PDUs from the input module 1 and processes them similarly.

Improved permeability layer mechanism

2 is a schematic representation of a method for transmitting soft information from a protocol layer L-1 to a protocol layer L + 1 via a transparent protocol layer L. FIG. The decoder module shown in FIG. 1 may implement processing in the transparent protocol layer L. FIG.

In the method of FIG. 2, protocol layer L receives a sequence of soft information (100), which includes the PDU of layer L11. For example, the soft information 100 sequence is provided by a lower protocol layer L-1 that obtains by encapsulating one or more PDU payloads of layer L-1. In this embodiment, the control field of the PDU 11 is a header 12 and the error detection code field is a trailer 13 containing a CRC or checksum associated with the header. In FIG. 2, 'n' is an index indicating a currently processed PDU, and 'n-1' refers to a previously processed PDU. The header 12 and the trailer 13 are converted into hard information and perform a CRC check to detect whether there is an error in the received header. If there is no error, the PDU is processed as a function of the information contained in the header 12. As a result, the payload 14 of the PDU is decapsulated and delivered as soft information to a predetermined destination, i.e., protocol layer L + 1.

If the CRC check shows an error, a header decompression step 15 is performed. The header decompression step includes two main principles. First, intra- and / or inter-layer redundancy is used to construct some a priori information on the candidate values for the header. Arrows 16 and 17 represent various redundant sources, which can be used to facilitate header recovery in layer L. Second, the CRC or checksum of the PDU 11 is used as an error correction code to select an optimal matched corrected value. Thus, the header decompression step combines the soft information provided by the lower protocol layer with the CRC or checksum characteristic and previously introduced a-priori information. The header 12 is removed after decoding, but for further processing, the information field is needed to transfer the payload 14 to the layer L + 1. In the illustrated embodiment, the payload 14 includes a PDU 18 of layer (L + 1).

For robust header recovery MAP Estimator

Next, a general model of the PDU, e.g., packet and corresponding header recovery methods, will be described. This model makes it possible to implement the method of FIG. 2 using any given protocol.

In a given layer L, the incoming packet includes one or more control fields, e.g., a header, one or more service fields, e.g., a payload, and an error detection field, e.g., a CRC. The information protected by CRC C of lc can be divided into four parts. The constant field represented by the vector k of lk bits is assumed to be known, e.g., a constant value of the protocol specification. The predictable field is included in vector p of lp bits. Unlike known fields, the predictable field is estimated by using intra-hierarchy and intra-hierarchy redundancy represented by R. [ These are predicted using the information contained in the previously received packet. The predictable field is assumed to globally determine whether the preceding packet was received correctly. An important unknown field is collected in vector u of lu bits. These parameters are either completely unknown or limited to a finite population of values? ( K , p , R), which may be a function of the values of k , p and R. Finally, the vector o of the ℓo bit contains any other fields covered by the CRC. This final part contains the unknown data, which is not required for processing of the packet in layer L but may be important in layer L + 1.

These fields are collected in a vector r = [ k , p , u , o ] of lr = lk + lp + lu + lo bits. The order of the bits of the vector r may not correspond to the order in which the data is transmitted in the packet.

CRC c associated to r is evaluated are as c = F (r), where F is a generic encoding function (generic encoding function). More specifically, the value of c depends on a generator polynomial that characterizes the CRC.

는 g(x)와 관련될 수 있다. G를 사용하여, c는 다음과 같이 반복적으로 얻어질 수 있다.Symmetric generator matrix

Can be related to g (x). Using G , c can be obtained repeatedly as follows.

Here, 'i' is an index indicating the number of repetitions.

의 j 번째 라인에 반드시 대응하는 것은 아니다. ℓr 반복 이후, c ^ℓr=F(r)이다.In equation (1), r ⁱ = [r ₁ ... r _i , 0 ... 0] and (+) is an XOR operator. ? (i) represents a parity vector associated with bit ri

Th line in Fig. After ℓr iteration, c ^ℓr = F ( r ).

It is also assumed that data is transmitted through an Average White Gaussian Noise channel introducing noise of zero mean and variance? ² . Noisy data and CRC from layer L-1 are displayed as follows.

This includes observations of k , p , u , o, and c . Since k and p can be predicted or predicted, only u remains and is estimated. Considering the CRC characteristic useful for estimating the information of observations y , k and p , the overlap R and u , the maximum inductive predictor (MAP predictor) is developed,

는 조건부 확률을 표시한다. 몇 번의 파생(derivations) 후, 다음을 얻는다.Here,

Indicates the conditional probability. After several derivations, you get:

Channels are less memory. Assuming that o is independent of R, we can write

의 연역적 확률을 나타내는데, 여기서 N(.,.)은 가우스 법칙을 표시하는데 제 1 인수는 평균 값이고 제 2 인수는 분산이다. In formula (5), P (u | k, p, R) , and u is

Where N (.,.) Denotes the Gauss law, where the first argument is the mean value and the second argument is the variance.

The possible values of u, i.e., the values held as candidate values,

Lt; / RTI >

Assuming that the combination of u is the same,

.

marginalizing equation (5) for a combination of 2 ^** ℓo of o ,

The characteristics of the CRC

.

Finally, when the equation (6) is combined with the equation (3), the MAP estimator is expressed as follows.

here

Represents the peripheral likelihood of the vector [ k , p , u ] as a function of the observation of error detection code y _c and other data y _o .

The categorization of the aforementioned fields is intended to include most transport protocols. Some field categories may not apply to specific protocols under consideration. That is, in a particular example, a portion of a field category is considered empty, for example, a known field k , a predictable field p, and / or other field o .

MAP Estimator  Second Example

In the preferred embodiment, the CRC covers only the control data, i.e., the header, in layer L. After that, o becomes empty and the sum of Eq. (6)

.

In this case, equation (7)

.

In formula (8), [k, p , u] vectors represent a candidate value set for the control data having an error, F ([k, p, u]) vectors error detection according to the candidate values of the set Code value, which can be directly evaluated by equation (1), e.g., a basic CRC calculation. Since the sum of all possible values of the vector o is not needed in that case, the computational complexity is greatly reduced compared to Eq. (7). Vectors k and p refer to one or more fields of a fixed header for a given PDU and u refers to one or more fields that may be for a number of values, i. The conditional probability of equation (7) is a function of the correlation function, that is, the correlation between the observed value of the unknown field for the first and each candidate value for the unknown field, and the error detection code field of the PDU for the control data. Can be evaluated by calculating the correlation between the observed value and the error detection code value associated with each candidate value.

This correlation can be calculated as the Euclidean distance between hard decision data. When the observation y is provided as soft data, the soft channel decoder determines the likelihood parameter for each possible bit state, i. E., The likelihood parameter with bit '1' and the likelihood parameter with bit '0' . The correlation between the hard data of the candidate values and the soft data of the observations can be computed as the incremental Euclidian distance, where the likelihood parameter is used as the weighting factor.

In general, solving Equation (8) allows to select one corrected value for the control data between the set of candidate values. In an embodiment, a transmission error affecting only the error detection code may be corrected by this method.

In equation (8), the known and predictable fields are not considered in the first conditional probability. However, it can be considered in a modified embodiment. Since these fields have a single value, they do not contribute in any way to distinguishing between the candidate values for the control data.

As described above, a deterministic relationship rule between a given layer, that is, an intra-hierarchical relationship or a field at a different hierarchy, i.e., an intra-hierarchical relationship, helps to recover an error header, Because.

Applying to the 802.11 standard

An embodiment of an IEEE 802.11 (WiFi) network will be described in which a header recovery method is implemented in the PHY layer and / or the MAC layer.

Figure 3 shows an example of an RTP / UDP / IP protocol stack, which can be used for multimedia packet transmission in combination with the 802.11 standard (WiFi). 'H' refers to the header of each PDU and APL refers to the application layer. At the PHY layer, the CRC protects the header field. At the MAC layer, the CRC protects the header and payload. In the IPv4 layer, the header field is protected by a checksum. At the UDP layer, the checksum protects the header and payload.

The 802.11 standard can be combined with other protocols. The protocol stack of Figure 3 is not limiting. The segments and encapsulation mechanisms shown in each protocol layer of Fig. 3 are merely illustrative and not limiting.

Figure 4 illustrates an embodiment of a network architecture, wherein the protocol stack of Figure 3 may be used. In this embodiment, consider the downlink multimedia transmission over the 802.11 air interface between the access point AP and the terminal T1. The multimedia stream 25 comes from the media server 20, for example via the broadband network 21 and the router R1.

5 is a schematic representation of a receiver device 30 included in a terminal T1 that receives and decodes a media stream. The receiver device 30 includes a wireless front-end 3! And a series of decoder modules 32-37 corresponding to each protocol layer. The wireless front-end 31 demodulates the received radio signal and conveys the baseband signal to the PHY decoder module 32 in the form of soft information.

The PHY and MAC layer specifications are now briefly described with reference to Figures 6 and 7.

DSS PHY  Layer Description

At the PHY layer, the 802.11 standard uses either Frequency Hopping Spread Spectrum (FHSS) or Direct Sequence Spread Spectrum (DSSS) to provide a 1 or 2 Mbps transmission rate in the 2.4 GHz band. In DSSS, an 11 chip breaker code sequence is used to spread a 1 Mbps bitstream. Thus, the coded flow represents an 11 MHz baseband signal. Differential BPSK (DPSKK) or Differential QPSK (DQPSK) modulation is applied to provide a bit rate of either 1 Mbps or 2 Mbps, respectively.

The DSSS PHY packet format 40 is shown in FIG. The preamble 41 and header 42 are transmitted using 1 Mbps DBPSK modulation and the payload 43 is modulated at 1 Mbps DBPSK or 2 Mbps DQPSK. In this PHY packet, the SYNC field 44 and SFD field 45 consist of 144 known bits, which are not protected by CRC. These fields can be used to estimate the noise variation induced by the channel.

The two-byte CCITT CRC-16 field 46 protects the signal field 47, the service field 48 and the length field 49. Using the generic model notation described above, the field 46 will be denoted by c ^PHY and its associated encoding function denoted F ^PHY . In this example, the payload 43, which is assumed to contain only one MAC packet, is not protected in this layer. The service field 48 is reserved for future reference. It is set to 00 ₁₆ according to the notation of the generic model and is included in k ^PHY . The value of the signal field 47 indicates the payload modulation and can be the same for OA ₁₆ or 14 ₁₆ only for 1 or 2 Mbps bit rates, respectively. The value of the length field 49 indicates the number of 2-byte microseconds required to transmit the payload 43. This depends on both the bit rate and the payload size. Which may range from 16 to (2 ^** 16-1). Using the general model notation described above, signal 47 and length 49 form the vector u ^PHY . In this layer, vector p ^PHY and o ^PHY are empty.

MAC  Layer Description

The MAC packet format 50 is shown in FIG. In this packet, the 4-byte CRC field 51 is denoted c ^MAC and protects both the header field 52 and the payload 53. The encoding function is denoted by F ^MAC .

Considering the non-encrypted downlink transmission of an aligned MAC data packet with a particular communication context, e.g., deactivated retransmission and power = save mode, all subfields of the 2-byte frame control field 54 are known or easily predicted , Except for more flags 55 that will change more dynamically. The 6 byte address field 56 is known because it contains the MAC address of the terminal T1. The final field 57 of the MAC header is reserved for the local wireless network. It consists of 6 bytes of zero when not in use. Using the generic model notation, all previously mentioned fields can be included in k ^MAC or p ^MAC .

The six byte address field 58 contains the MAC address of the access point (AP). This address will be transmitted to the terminal T1 during the media retention procedure (RTS-CTS) and will remain known by the receiver during the entire communication session. The 6-byte address field 59 corresponds to the MAC address of the router R1. As soon as the access point (AP) has already received the router address of one information packet, the address field (59) can be easily predicted by the receiver for subsequent packets if the access point (AP) is connected to a single router.

The 2-byte subsequent control field 60 contains two parameters, a sequence number and a fragment number. The sequence number indicates the value of the current IP packet counter. The fragment number indicates the value of the current MAC packet counter. Assuming that the packets are transmitted in order, these parameters can be easily determined, the sequence number is incremented by one for each RTS-CTS and the fragment number is incremented by one for each received MAC packet. Thus, the receiver can estimate the sequence control field 60 by increasing the previously received value. All these predictable single value fields can be treated as belonging to p ^MAC .

More flag flags 55 specify whether the current MAC packet is the last for the fragment sequence of the IP packet. The two bytes of the duration field 61 indicate the number of microseconds required to transmit the next MAC packet and some control parameters. The value depends on the current modulation and the size of the next MAC packet. These two fields are included in the vector u ^MAC of the general model. Finally, the payload contains the data to be transmitted and its size is between 0 and 2312 bytes. This is expressed as o ^MAC .

In one embodiment, in order to allow some packet loss and / or a certain degree of packet realignment, the receiver may set the sequence control field < RTI ID = 0.0 >Lt; RTI ID = 0.0 > 60 < / RTI > This field is then p ^MAC Instead, it should be considered in the vector u ^MAC .

In one embodiment, to allow multiple routers, the receiver may configure a limited set of candidate values for the address field 59 by accessing the most recent or most frequent list of values. This field is then p ^MAC Instead, it should be considered in the vector u ^MAC .

Hierarchical correlation

To illustrate the correlation between the 802.11 PHY and MAC layer, transactions at the MAC layer will be described. FIG. 8 shows an 802.11 MAC transmission protocol when two MAC data packets are to be transmitted. The transmission is initiated by a media retention procedure that constitutes an RTS-CTS exchange between the access point (AP) and the terminal (T1). Thereafter, a data packet is transmitted to the terminal T1, which acknowledges the packet reception (ACK). It is assumed that control packets such as RTS, CTS and ACK are correctly received by the terminal T1. Only errors in the data packet will be considered. A short inter-frame space (SIFS) of 10 μs separates each packet to prevent collision. When all the packets are received by the terminal T1, the transmission is terminated and the DIFS (Distributed Inter-Frame Space) of 50 μs proceeds to the next media preservation procedure. The duration field 61 is included in each packet and its value indicates the number of microseconds required to transmit the next packet. The duration value allows to adjust the network assignment vector (NAV) for the other terminal. Other stations can not communicate during the NAV period to prevent interference.

Next, D ^MAC and B ^PHY are set to a value of the transmission bit rate coded in the signal field 47 associated with the value of the duration field 61 and the nth packet (RTS or data packet) transmitted by the access point (AP) Value. According to the MAC layer specification of the 802.11 standard, D ^MAC is defined as follows.

Except for the last packet of the fragment sequence of an IP packet when the value of the more flag MMAC is zero. in this case,

.

는

비트의 다음 PHY 페이로드의 전송 기간을 지칭한다.In Equations (12) and (13), T _SIFS denotes a period of SIFS, and T _OVH denotes a period of transmitting PHY overhead having a limited size composed of a preamble and a header at 1 Mbps. The other term in equation (12) depends on the current bit rate B ^PHY _n . CTS and ACK have the same limited size? _CA , and? _CA / B ^PHY _n corresponds to a period of transmitting one of these packets. Finally,

The

Quot; refers to the transmission period of the next PHY payload of the bit.

PHY  Restore header

For a given packet 'n' in the PHY layer, the observations associated with k ^PHY _n , u ^PHY _n and c ^PHY _n defined above are

Lt; / RTI >

Also, y ^PHY _{x , n} represents observations related to l ^PHY _{x , n} bits of the payload. The number of values that the u ^PHY will take is significantly reduced when using the duration field contained in a previously received MAC packet (RTS or data packet). In ^PHY B _{n -1} and MAC layers for the previous packet of the PHY layer using a D ^MAC _{n -1} for the previous packet, it is possible to deduce ℓ ^PHY _{x, n,} as follows from the formula (12).

Thereafter, the coded period L ^PHY _n in the length field 49 of the current PHY packet n may be calculated using:

In equation (14), l ^PHY _{x , n} is completely determined assuming a correct estimate of the header of previous packet 'n-1' at the PHY and MAC layer. Thereafter, according to equation (15), L ^PHY _n can take only two values depending on B ^PHY . Integrating these structural correlations in Eq. (8) yields a definition of the MAP estimator specific to the PHY layer.

Therefore, only the values of the signal 47 and length 49 fields that satisfy Eq. (15) are stored in? ^PHY _{u, n} ( k , p , R) and are considered for correcting the header.

The intra-layer and intra-layer redundancy described above can be used in the enhanced PHY layer decoder module 32 similarly configured to the decoder module 10 of FIG. In this embodiment, the control data correction module 6 accesses the information about the previously received packet to determine the value of the length field 49, which satisfies equation (15). For example, the control data correction module 6 computes l ^PHY _{x , n} via equation (14) by reading the parameter B ^PHY _{n -1} of the caching module 5, which is processed by the processing module 3 It was previously recorded. Similarly, the parameter D ^MAC _{n -1} may be obtained from the MAC decoder module 33, for example, via temporary data storage.

In this embodiment, the control data correction module 6 constructs a set of two arbor values for the PHY header 42 of packet 'n', which has two values of the signal field 47 allowed by the protocol specification , I.e., OA ₁₆ and 14 ₁₆ only. The final CRC value for field 46 is also calculated. The control data correction module 6 then evaluates Eq. (16) to select candidate values that best match the observations, i. E., The header field 42 and the received values of the CRC field 46. [ This allows you to restore the correct header value in many cases. The processing module 3 of this embodiment then uses the recovered value of the length field 49 to correctly range and decapsulate the payload of the PHY packet 'n'.

The above-described embodiment assumes that the service field 48 is set to be constant at 00 ₁₆ . However, if a particular implementation of the protocol allows multiple values for the service field 48, the corresponding PHY layer decoder module may be implemented based on the same principle. To configure the candidate value set for header recovery, all possible values of the service field 48 are considered in combination with the two possible values of the signal field 47.

FIG. 9 shows an embodiment of a decoding method for a PHY layer, which may be executed by the PHY layer decoder module 32 at a receiver. Arrow 62 shows the exchange of information between layers and between successive packets. The header restoring step 66 is performed as described above.

Y ^PHY _{s , n} represents the observation of a known preamble s ^PHY of l _s ^PHY bits. As described above, receiver synchronization is performed using s ^PHY . In one embodiment, the noise power sigma ² is estimated from s ^PHY and y ^PHY _{s , n} . This measurement is useful for operating with soft information because it can evaluate all possibilities. The noise power estimator

는 유클리드 거리를 표시한다.Lt; RTI ID = 0.0 >

Indicates the Euclidean distance.

Bypassing the header decompression step (66) when a normal CRC check is successful can minimize computational complexity. In one embodiment, the header decompression processing is deactivated when the quality of the soft information provided by the lower layer is too poor, i.e., when the signal power is below a predefined threshold. In this case, the packet is discarded.

In FIG. 9, the MAC decoding step 63 may or may not include the MAC header decompression method.

PHY  Simulation result

The transmission system 64 shown in Fig. 4 consisting of a transmitter AP, an AWGN channel 65 and a receiver Tl has been numerically simulated. The AP generates PHY and MAC packets according to the format defined in FIG. 6 and FIG. The MAC payload consists of a variable amount of randomly generated bytes. The transmitter modulates the data in DBPSK for all simulations. Three types of PHY header recovery methods are considered in T1. The standard decoder performs a hard desizing of the data at the channel output. The robust decoder uses only the intra-layer and inter-layer redundancy through the soft decoding algorithm, leaving the information provided by the CRC. Finally, the CRC robust decoder uses Equation (16) to combine intra- and inter-layer redundancies with information raised by the CRC through the header decompression method provided in the previous section.

Performance analysis in terms of erroneous header rate (EHR) versus signal-to-noise ratio is provided in FIG. In FIG. 10, a standard, robust, CRC-robust PHY decoder is compared. Rugged decoders outperform standard decoders. An EHR of less than 10-5 is obtained for a robust decoder for SNR above 4 dB and for a robust decoder and CRC-robust decoder for SNR above 2 dB. Using a standard decoder, a SNR of at least 15 dB is required to obtain a comparable EHR. At the PHY layer, considerable coding gain is observed for relatively low additional complexity.

MAC - Light header restoration

In the embodiment of the transmission system of FIG. 4, the MAC layer is implemented in a modified packet format, wherein the CRC field 51 is modified to protect only the header field 52 of the MAC packet. This modified MAC protocol will be referred to as MAC-Lite, in contrast to the UDP-Write protocol described in Request For Comment 3828 by the Internet Engineering Task Force. In this case, since the vector o ^MAC is empty, the MAC-write decoder module can implement a dae recovery method using a MAP predictor defined as follows.

here,

Thus, at the MAC-Lite layer, the header decompression method can be used with steps similar to the PHY layer described above. To construct a candidate value, a known predictable field defined in the MAC layer description is set to each single predicted value. In this embodiment, the MAC-write decoder module uses these in-layer redundancy to determine these values.

The PHY packet payload y ^PHY _{x , n} includes observations that enter the MAC layer and are associated with k ^MAC _n , p ^MAC _n , u ^MAC _n , o ^MAC _n, and c ^MAC _n specified in the MAC RP layer description. Using the conventional method defined in the general model, the following can be used.

의 가능한 값은 다음에 의해 주어진다.The number of possible combinations for u ^MAC _n can be greatly reduced when using the properties of equations (12) and (13). In fact, the value of period D ^MAC _n is determined wholly when M ^MAC _n = 0. If M ^MAC _n = 1, the duration field 61 depends on the next PHY payload size. The number of combinations is a function of the range of the MAC payload size. Considering that the payload contains the total number of bytes, Equation (12)

The possible values of < RTI ID = 0.0 >

Where i = 1, 2 ... 2312.

In equation (17), ℓ _HDR specifies the known size of the header of the MAC data packet. Then, using equations (12), (13) and (17), u ^MAC is limited to 2313 combinations, which are inserted in the set of candidate values Ω ^MAC _{u , n} .

A standard, robust, and CRC-robust MAC-write decoder is compared in FIG. Three types of MAC-light header recovery methods are considered in T1. The unfolded decoder performs a hard desynthesis for the data provided by the PHY layer. A robust decoder uses only intra-layer redundancy through a soft decoding algorithm, leaving the information provided by the CRC. Finally, the CRC robust decoder combines the redundancy in the hierarchy with the information provided by the CRC via the header decompression method based on equation (20). EHR is lower than 10 ^-5 for SNR greater than 3 dB when using CRC redundancy, and two other methods require at least 14 dB.

Each of the above-described robust decoder modules in the PHY layer and MAC-write layer can be used independently or in combination with each other. In a combined embodiment, the decoder modules 32 and 33 of FIG. 5 implement a corresponding header decompression method, which increases the number of payloads reaching the next protocol layer. In a numerical simulation, the combination of the proposed transmissive PHY and MAC-light layer mechanisms eventually restored all PHY and MAC headers since 3dB SNR.

In the embodiments of the above-described PHY and MAC layer decoder module, assigning each header field to a category of a model defined by a known vector k , a predictable vector p, and an unknown vector u is only exemplary and not restrictive. This assignment is based on the assumption that some elements of the communication context are fixed deductive. For other uses, there may be a known deductive argument of more or less context information. As a result, in other embodiments, some of the fields provided as being known or predictable will be treated instead as unknown. The computational complexity of the header reconstruction process may increase as more fields are treated as unknown, but appropriate use within or across layers will allow this complexity to be maintained at acceptable levels in many instances.

MAP Estimator  Second Example

In other protocols, the CRC covers not only the control data at layer L but also some or all of the service data, e.g., the payload of the canonical 802.11 MAC layer. In a general model, this situation implies that the vector o is not empty. An embodiment of a MAP estimator suitable for restoring the control data of the PDU, in this case the MAC packet header will be described.

If o is not empty, it is assumed that the bits of o are iid and do not depend on other parameters. The second term in Eq. (6) is as follows.

및

을 정의하는데, 여기서 N(.,.)은 가우스 법칙을 표시하고 제 1 인수는 평균 값이고 제 2 인수는 분산이다.In equation (9), p (o) is the a-priori probability of the vector o ,

And

, Where N (.,.) Denotes the Gauss law, the first argument is the mean value and the second argument is the variance.

Evaluating equation (9) requires calculation of the sum of the probabilities associated with the combination of 2 ^** lto of the vector o and the corresponding CRC. This evaluation has an obviously complex index of ℓo. In this section, two methods are proposed to reduce the complexity of Eq. (9). The first contains the correct calculation and the second provides the approximate solution.

Accurate sum calculation

The method includes a trellis configuration consisting of a grouping combination of values taken by o to obtain the same value of CRC. The trellis consists of a set of states interconnected by segments. This state is grouped into a set of tO + 1 indexed by a parameter I = 0, 1 ... l0 representing the depth and corresponding o bits of the trellis. At any depth j, there is a 2 ^** l _c state, i. E., There is a 2 ^** l _c value of the CRC. The segment connects the states between the depths j and j + 1, where j = 0,1 ... ℓo-1. At the depth j, each state indexed with i = o, 1 ... 2 ^** tc-1 takes a value S _i (i). The index i specifies the numerical value of the CRC _ci . Thus, the vector c _i contains a binary representation of i.

The step of evaluating equation (9) will be described later. The trellis starts at depth 0 and proceeds to depth lo.

Step 1 - At depth j = 0, Si (0) = 0 and i = u and Su (o) = 1 for all i are the exception. The only possible state at depth 0 is determined by c _u = F ([ k , p , u , 0]), which represents the value of the CRC when vectors k , p and u are known.

Step 2 - For j = 1, 2 ... < RTI ID = 0.0 > l0, < / RTI > each state is connected to two previous states depending on the jth entry of vector o.

로 표현되는 정수이다.i = 0, 1 ... 2 ** lc-1 and q is

Lt; / RTI >

Step 3 - After the ℓo iteration, we obtain any i = 0,1 ... 2 ^** tc-1.

Multiplying Si (? O) by P (y _c | c _i )

L _i depends on all values of o providing the same CRC _ci .

Step 4 - By summing the L _i values for i = 0,1 ... 2 ^** tc-1,

.

Example 1: Figure 12 shows a trellis, which can be obtained using the following for a symmetric binary thaw (7,4) code.

In this example, assume that [ k , p , u ] = [1] and lo = 3.

For a given value of k, p, and u , this technique allows reduction of the complexity for the evaluation of Eq. (9) from O (2 ^** ℓo) to O (ℓo2 ^** ℓc). Nevertheless, the initialization in step 1 depends on the value taken by F ([k, p, u]). For all values of u in Ωu, a new grid must be constructed. Thus, the overall complexity is O (| Ωu | ℓ2 ^** ℓc) and | Ω | represents the cardinal number of Ωu.

In the following, a method of simultaneously evaluating Eq. (9) for all initial states at depth 0 (2 ^** tc possible values of CRC) is proposed. The method includes constructing a trellis by backing up to a depth of 0, starting from a depth of l0. The steps for the sum evaluation are provided as follows.

A | _{(c i} y c) - stage 1 depth j = ℓo, any ^{ii = 0,1 ... 2 ** Si (} ℓo) for ℓc-1 = P.

Step 2 - For j = ℓo-1 ... 1,0

의 수치 값을 나타낸다. 깊이 j에서, 상태 i는 상태 ii에 접속되고 깊이 j+1에서 q이다.For all i, q is

. At depth j, state i is connected to state ii and q at depth j + 1.

Step 3 - After repeating the loop, for i-u,

cu = F ([k, p, u, 0]).

Example 2: Using the same code as in Example 1, a trellis (composed of the rear) shown in FIG. 13 is obtained. This method makes it possible to directly calculate the sum of equations (0) for all values taken by u with complexity O (ℓo2 ^** ℓc).

Approximate sum calculation

Most CRCs are larger than 16 bits. In some embodiments, the complexity O (? O2 ^**? C) may be too large to allow real-time implementation of the above-described method. The approximate calculation consists of dividing the CRC into ℓb = ℓc / n _b bits of the n _b block. For example, CRC with length ℓc = 32 can length can be divided into n _b = 4 blocks with ℓb = 8. Each of these blocks is then assumed to be statistically independent of each other. Using this decomposition, y _c can be written as:

Using an independent assumption, equation (9)

Respectively,

, Where Fm is an encoding function associated with column m-1. The lb + 1 to m.lb of the scan evaluates the partial CRC of lb bits. The evaluation of equation (11) is similar to that of Ψ described in the previous section. The only difference is the size of the trellis: 2 ^** lb should be considered at any depth (instead of 2 ^** lc without dividing the CRC). The total computational complexity to evaluate Eq. (10) is now O (n _b ℓo2 ** (ℓc / n _b ).

The second embodiment of the MAP estimator allows restoration of transmission errors in a field of interest identified by a selected portion of the information covered by the error detection code, i.e., u, k and p, Ignoring any potential error affecting the other part of the information, i.e., the auxiliary field identified as o. A transmissive protocol layer decoder can be implemented that restores errors of only control data that this type of MAP estimator functions. This type of embodiment may be used for a standard 802.11 MAC layer. The decoder module 10 shown in Fig. 1 can be easily configured to implement this embodiment. The function of the control data correction module 6 is modified to allow determination of the likelihood of each candidate value of the header.

MAC  Restore header

Combining the characteristics of the 802.11 MAC layer of Equation (7), the following MAC metric is obtained:

Here, the second term can be calculated by the method provided in the second embodiment of the MAP estimator, expressing the likelihood of each candidate value of the header.

A robust MAC decoder that implements a header decompression method based on Eq. (18) has been numerically simulated under conditions similar to PHY layer simulations. Figure 14 compares the coding gains obtained by the standard, robust, and CRC-robust MAC decoders. Two payload sizes (50 and 100 bytes) were considered. In addition, the approximate sum calculation method described above is used, and the CRC is divided into four blocks of one byte each. The shape of the curve is very similar to the result obtained at the PHY layer, but the coding gain is significantly less. The gain due to the MAC CRC information improves as the SNR increases. Using a payload of 100 bytes, an EHR of less than 10 < ^-5 > is achieved for SNRs of 11 dB, 14 dB and 15 dB, respectively, when using CRC0 robust, robust and standard MAC decoders. MAC header recovery processing is more complicated than that performed at the PHY layer due to the marginalization operation required in Equation (9). The larger the payload, the more complex the decoding process. The difference between a robust MAC decoder and a CRC-robust MAC decoder is that it considers the last term of Eq. (9), i.e., the correlation of CRC values. Figure 14 makes it possible to recognize the contribution of this feature to the efficiency of CRC-robust MAC decoders.

Figure 17 is a representation similar to Figure 9, which is an embodiment of a decoding method suitable for an 802.11 protocol stack. Figure 17 illustrates the steps of PHY header decompression 85 based on evaluating Equation 16 that may be performed by PHY layer decoder module 32 and the steps of PHY header decompression 85 that may be performed by MAC layer decoder module 33 at the receiver Combines the steps of MAC header recovery 86 based on evaluating equation (18). With this combined embodiment, a high number of MAC payloads, for example, in the form of soft dissociation data, as indicated by reference numeral 87, can be delivered to the upper layer.

The above-described second embodiment of the MAP estimator can be applied to any other application where selective recovery of transmission errors is required in addition to the transparent protocol layer.

In one embodiment, the second embodiment of the MAP estimator serves as the basis for an error correction method applied to error resilient video transmission using a scalable video codec. A scalable video codec is defined as a codec capable of generating a bitstream that can be divided into an embedded subset. These subsets can be independently decoded to provide an increasingly quality video sequence. Thus, a single compression operation can produce a bitstream with a different rate and a corresponding reconstructed quality level. Scalable video codecs are supported by most video compression standards such as MPEG-2, MPEG-4 and H.263.

Video data partitioning can be used to help scale possibilities. For example, in MPEG-2, the slice layer indicates the maximum number of block transform coefficients included in a particular bitstream, known as a priority breakpoint. Data segmentation is a frequency domain technique that divides 64 quantization transform coefficients into two bit streams. First, the higher priority bitstream, e.g., the base layer, includes more significant lower frequency coefficients and side information, e.g., zero frequency values and motion vectors. Second, the lower priority bitstream, e.g., the enhancement layer, conveys higher frequency video data.

In the embodiment shown in FIG. 15, a subset of the original bitstream is transmitted as key data to provide base layer quality in one or more of the overlay layers and transmitted as ancillary data to provide an enhancement layer. Thus, the error correction method allows heterogeneous error protection, e.g., allowing more robust error protection to the base layer than the enhancement layer and transmitting multiple layers to the same packet. As a result, the base layer can be successfully decoded with a higher probability even during negative transmission channel conditions.

15 is a schematic representation of a transmission system 80 that transmits video, i.e., an oil sequence. The system 80 includes a transmitter device 70 that transmits video data to a receiver device 71 on a communications channel 72, e.g., a wireless channel. This video is encoded with a scalable video codec. In the transmitter device 70, the coder module 73 encodes a picture sequence with a scalable video codec to generate a first bitstream corresponding to the base layer and at least one second bitstream corresponding to the enhancement layer, And places successive segments of the bitstream in the transmitted packet 74. Thus, the coder module 74 generates a data packet sequence 74. In the illustrated embodiment, the packet 74 includes a main field 75 for data belonging to the first bit stream, an auxiliary field 76 for data belonging to the second bit stream, And an error detection field 77 for an error detection code (e.g., CRC). The data packet 74 sequence is transmitted over the communication channel 72 using any suitable transport protocol or protocol stack and any suitable encapsulation technique. Thus, in the transmitter device 70, the coder module 73 can be connected to an additional coder module corresponding to the lower protocol layer, which is not shown in Fig.

In the receiver device 70, the decoder module 90 receives successive packets 74, for example, from additional decoder modules corresponding to lower protocol layers not shown. Decoder module 90 checks the CRC received in field 77 for the corresponding data contained in the packet. If the CRC check is unsuccessful, the decoder module 90 performs an optional error recovery method to recover the main field 75, e.g., transmission errors affecting the base layer.

FIG. 16 is a schematic representation of an embodiment of decoder module 90. FIG. The decoder module 90 includes an input module 91 for receiving the data packet 74 and an error detection code checking module 92 for checking the error detection code of the received packet. If the check is successful, the module 92 removes the field 77 and passes the video data of the fields 75 and 76 to the output module 93. The output module 93 outputs the video data to a suitable video decoder, for example, MPEG-4, which decodes the first and second bit streams to reconstruct the picture sequence.

If the check is successful, the error detection code checking module 92 sends the packet 74 to the selective error correction module 94, which operates to find the most correct value of the data belonging to the first bit stream, The main field 75 is restored. The optional error correction module 94 determines an exclusive list of candidate value finite sets for the main field 75, e.g., 2 ^** L binary values, where the a priori known information allows excluding some of these binary values L is the length of the field 75 or more limiting set. For each set of candidate values, the optional error correction module 94 may be used as a function of the data received in the fields 76 and 77 of the received packet 74, e.g., by evaluating equations (9) and (10) Determines the perimeter likelihood of candidate values. The optional error correction module 94 also determines the Euclidean distance between the data and the candidate value received in the correlated, e.g., the main field 75 of the received packet 74. It then uses each of the surrounding likelihood and correlation values to select the candidate values for the main field, which best matches the observations. The optional error correction module 94 then passes the corrected value of the main field 75 to the output module 93 so that at least the base layer can be decoded. The received value of the auxiliary field 76 may also be passed to the output module 93, which will not be corrected by the process described above if there is a transmission error affecting it.

In a modified embodiment of the transmission system 80, the packet 74 further includes a header field 78, which is also covered by the error detection code in field 77. For example, such packet 74 may be used in a UDP or UDP-light protocol layer. The optional error correction module 94 may then recover the transmission error found in the main field 75 in addition to the header field 78. [ Assuming that these fields are independent, the candidate values to be considered in this case include any combination of candidate values for the control data of the header field 78 having a candidate value for the video data of the main field 75.

In another embodiment, the packet 74 may comprise a number of primary and / or secondary fields. For a decoder module 90 that processes primary and ancillary data, a corresponding map of the packet format must be available to the decoder module 9. Which may be configured to be available to the decoder module in any suitable manner, e.g., fixedly configured or communicated over another channel. Preferably, the main field or field is located closer to the packet head than the auxiliary field, but this is not necessarily so.

The above-described methods and apparatus for recovering transmission errors may be applied to other transport protocol layers, such as DVB, Wimax, and non-standard protocol layers, as described above. A restoration method and apparatus for restoring transmission errors can be used in a single protocol layer or in various protocol layers of a protocol stack.

The above-described methods and apparatus for recovering transmission errors can be applied to the transmission of any kind of data. It is particularly contemplated to apply the transmission of compressed multimedia content because the payload of the PDU in any lower layer may contain a large amount of application data, i.e., multimedia content information. Application to the transmission of information over wireless channels is especially taken into account, since they are vulnerable to transmission errors. Broadcast transmissions, such as transmission to satellite television, or transmissions with strong delay constraints, such as video telephony, are particularly contemplated because they are difficult to rely on retransmissions to recover erroneous packets.

The above-described methods and apparatus for recovering transmission errors can be applied to a transparent protocol layer decoder, in particular for one or more lower transport protocol layers of a protocol stack, where the joint source-channel decoding technique has many errors at the application layer To be corrected. For example, by increasing the amount of packets reaching the application layer in the form of soft information, such a transparent protocol layer decoder can improve the performance of joint source-channel decoding at the application layer.

The functions of the various elements shown in the figures, including any functional blocks referred to as "modules ", may be provided through use of hardware capable of executing software associated with suitable software in addition to dedicated hardware. If provided by a processor, the functions may be provided by a single dedicated processor, by a single shared processor, or by a plurality of individual processors, some of which may be shared. Further, explicit use of the term " processor "or" controller "should not be considered as referring exclusively to hardware capable of executing software, and may be embodied by a digital signal processor (DSP), a network processor, an application specific integrated circuit (ASIC) , A field programmable gate array (FPGA), a read-only memory (ROM) that stores software, a random access memory (RAM), and a non-volatile storage device. Other hardware, conventional and / or conventional, may also be included.

The present invention is not limited to the above-described embodiments. The appended claims should be construed to encompass all such modifications and alternative constructions that may occur to those skilled in the art, which are expressly within the scope of the present invention.

In the claims, any element expressed as a means for performing a particular function includes any manner of performing the function, for example, a) a combination of circuit elements performing that function, or b) an rm function Firmware, microcode, etc., in combination with suitable circuitry for executing the software to perform. The invention defined by these claims is intended to combine and combine the functions provided by the various cited means in the manner claimed. Applicants therefore regard any means capable of providing these functions to be equivalent to those shown herein.

The use of the verb "comprise" and its use forms does not exclude the presence of elements or steps other than those stated in the claims. Also, the use of an article "one" preceding an element or step does not exclude the presence of a plurality of such elements or steps. The present invention can be implemented in hardware and software. The same hardware item may represent several means.

In the claims, any reference signs between parentheses shall not be construed as limiting the scope of the claims.

Claims (15)

CLAIMS 1. A method for recovering transmission errors,
Receiving the data packet including an error detection code (51, 77) associated with data contained in a data packet (50, 74), the data associated with the error detection code including primary data (52, 75) Comprising data (53, 76)
Checking an error detection code of the received packet to detect an error condition of the associated data;
If an error condition is detected, determining a finite set of candidate values for the key data;
For each candidate value of the set,
Determining a marginal likelihood of the candidate value via the assistance data as a function of an error detection code of the received packet;
Determining a first correlation between the primary data of the received packet and the candidate value,
Selecting a corrected value for the key data from the set of candidate values as a function of the marginal likelihood and the first correlation
A method for restoring transmission errors.
The method according to claim 1,
Wherein determining the peripheral likelihood of the candidate value comprises:
Determining a finite set of potential values for the ancillary data;
Classifying the potential value of the ancillary data into a subset associated with each error detection code value; all potential values of the ancillary data in the subset are calculated to produce the associated error detection code value when combined with the candidate values of the primary data Selected - and,
For each subset, determining a perimeter likelihood of the ancillary data belonging to the subset as a function of the ancillary data of the received packet and the error detection code of the received packet;
And determining a peripheral likelihood of the candidate value by accumulating the peripheral likelihood for all the subsets
A method for restoring transmission errors.
3. The method of claim 2,
Wherein determining the peripheral likelihood of the auxiliary data belonging to the subset comprises:
Determining a second correlation between an error detection code value associated with the subset and an error detection code of the received packet;
Determining a third correlation between a potential value of the ancillary data in the subset and ancillary data of the received packet;
Determining a peripheral likelihood of the auxiliary data belonging to the subset as a function of the second correlation and the third correlation
A method for restoring transmission errors.
The method according to claim 1,
Wherein determining the peripheral likelihood of the candidate value comprises:
Dividing the error detection code of the received packet into a plurality of blocks;
Determining, for each block of error detection codes of the received packet, a partial surrounding likelihood of the candidate value as a function of the block;
Determining a peripheral likelihood of the candidate value as a function of the partial peripheral likelihood associated with all blocks of the error detection code
A method for restoring transmission errors.
5. The method of claim 4,
Wherein determining the partial peripheral likelihood of the candidate value comprises:
Determining a finite set of potential values for the ancillary data;
Classifying the potential value of the ancillary data into a subset associated with each value of the block of error detection codes, wherein all potential values of the ancillary data in the subset are combined with the candidate values of the primary data, - < / RTI >
For each subset, determining a partial perimeter likelihood of ancillary data of the received packet and ancillary data belonging to the subset as a function of a block of error detection codes of the received packet;
And determining the partial peripheral likelihood of the candidate value by accumulating the partial peripheral likelihood for all the subsets
A method for restoring transmission errors.
6. The method of claim 5,
Wherein determining the partial peripheral availability of auxiliary data pertaining to the subset comprises:
Determining a second correlation between a block value associated with the subset and a block of error detection codes of the received packet;
Determining a third correlation between a potential value of ancillary data in the subset and ancillary data of the received packet;
Determining a partial periphery likelihood of ancillary data belonging to the subset as a function of the second correlation and the third correlation
A method for restoring transmission errors.
7. The method according to any one of claims 2 to 6,
Wherein classifying the potential value of the ancillary data into a respective error detection code value or a subset associated with each block value comprises configuring a trellis representation of the potential value of the ancillary data
A method for restoring transmission errors.
8. The method of claim 7,
The trellis representation consists of a low depth value to a high depth value,
The depth of the trellis represents the number of bits of the already determined auxiliary data.
A method for restoring transmission errors.
8. The method of claim 7,
The trellis representation consists of a high depth value to a low depth value,
The depth of the trellis represents the number of bits of the already determined auxiliary data.
A method for restoring transmission errors.
7. The method according to any one of claims 2 to 6,
The finite set of potential values for the ancillary data includes all possible binary value combinations for every every bit of the data of the field of the packet reserved for the ancillary data
A method for restoring transmission errors.
7. The method according to any one of claims 1 to 6,
The packet includes a protocol data unit (50)
Wherein the main data comprises control data (52) of the protocol data unit and the auxiliary data comprises service data (53) of the protocol data unit
A method for restoring transmission errors.
12. The method of claim 11,
Processing the service data (53) of the received protocol data unit as a function of the corrected value of the control data
A method for restoring transmission errors.
7. The method according to any one of claims 1 to 6,
The packet 74 includes video data coded with a scalable video codec,
The main data 75 comprises video data corresponding to a base layer and the auxiliary data 76 comprises video data corresponding to an enhancement layer
A method for restoring transmission errors.
An apparatus (90) for recovering transmission errors,
Input means (91) for receiving said data packet comprising an error detection code (77) associated with data (75, 76, 78) contained in a data packet (74) 75 and ancillary data 76,
An error detection code checking means (92) for checking an error detection code of the received packet to detect an error state of the associated data;
Determining a finite set of candidate values for the primary data and for each candidate value of the set a perceptibility of the candidate values via the secondary data as a function of the error detection code of the received packet, To determine a first correlation between the primary data of the received packet and the candidate value and to select a corrected value for the primary data from the set of candidate values as a function of the marginal availability and the first correlation Which includes the main data correction means 94
A device for restoring a transmission error.
14. A permeable protocol layer decoder comprising an apparatus and a processing module (93) according to claim 14,
The packet includes a protocol data unit (50)
Said main data comprising control data (52) of said protocol data unit, said auxiliary data comprising service data (53) of said protocol data unit, said processing module being operable to transmit said service data as a function of said control data Capable of operating to process
Transparent protocol layer decoder.

Images